Field of the Invention
The present invention relates to a network device, and a method of specifying data, for collecting data required for performing a malfunction analysis and a malfunction prediction from a device connected via a network, for example. In particular, the present invention relates to a network device, and a method of specifying data, for collecting data in order to perform a malfunction prediction efficiently, economizing communication and stored data capacity.
Description of the Related Art
Conventionally, maintenance systems in which a management system is installed on the Internet, and customer devices are managed via networks are known. In such conventional maintenance systems, device management is performed by the management system collecting information such as device statuses transmitted from a device monitoring apparatus, event data such as malfunctions, paper jams, or the like, the number of sheets in printing, or the like, for example. Also, in recent years, by collecting detailed device status data from a device monitoring apparatus, cause analysis and prediction of the occurrence of events such as malfunctions, jams, or the like has been performed. As a specific cause analysis, a data analyst, using table calculation software, a BI (Business Intelligence) tool, or the like, makes visible the causal relationship of device status data and the occurrence of events such as malfunctions, jams, or the like. Also, as prediction, an approach using regression analysis is common. A data analyst generates a malfunction analysis or malfunction prediction regression model based on device statuses, event data, and status data of a device that the management system collects. Malfunction analysis and malfunction prediction models are used for the purpose of application to serviceman maintenance plans, presentation of a warning to a user who is the owner of the device, or the like.
Meanwhile, in order to perform high precision analysis and prediction, it is necessary to generate an analysis model based on various additional information and not just simple status data of the device. In Japanese Patent Laid-Open No. 07-137358, improvement of prediction accuracy for a window time until a device malfunctions is realized by generating a model in accordance with device usage condition data in addition to device status data.
However, in Japanese Patent Laid-Open No. 07-137358, high precision prediction can only be performed with an analysis model in accordance with device status data and usage condition data. For this reason, newly generating an analysis model having a close relationship with, for example, device part information, a time over which the device is continuously turned on, or the like, cannot be performed, and it cannot be said to be highly versatile. Also, there is a problem in that even if configuration is taken so as to make collection of status data from the device to be high density and to be for various types in order to improve versatility, in such a case, it is necessary to accumulate a large amount of status data in the management system and so operation costs of the server increase.